Superalloys 718, 625, 706 and Derivatives 2005 Edited by E.A. Loria
TMS (The Minerals, Metals & Materials Society), 2005
Effect of
1
δ-phase on the Weldability and the Hot Ductility of Alloy 718
Göran Sjöberg, 2 Tomas Antonsson , 3 Saied Azadian , 3 Richard Warren and 2 Hans Fredriksson
,
Volvo Aero Corporation, 461 98 Trollhättan, Sweden
Royal Institute of Technology, 100 44 Stockholm, Sweden
Malmö Högskola, 205 06 Malmö, Sweden
Keywords: Alloy 718, weldability, δ-phase, hot ductility
Abstract
In the weld repair cycles of alloy 718 the solution temperature is usually selected low enough to
avoid grain growth in wrought material. At such low temperatures, however, δ-phase also
precipitates and the amount increases with each repair cycle.
It is well understood that the overall ductility is reduced by the presence of the δ-phase in alloy
718 and, as such, may reduce the weldability. But it is also reported that the δ-phase may cause
incipient melting in grain boundaries at the heat affected zone, HAZ, through a niobium rich
melt fed by the local abundance of this element from attached δ−platelets. The purpose of the
present work is to shed light on this issue on the one hand through practical weld experiments
under heavily constrained circumstances and on the other through high temperature ductility
testing in special hot tensile test equipment by which the hot ductility of re-melted material also
was possible to examine. Materials with two levels of δ-phase and with two grain sizes were
included.
Contrary to what was anticipated it was extremely difficult to provoke cracking even in
inhomogeneous cast material at the highest δ−levels in heavily constrained TIG-weld
experiments. Also the hot ductility test results, measured as the high temperature ductility
recovery temperature and range, indicated that there is no or very limited deterioration by the
presence of δ-phase. Anomalous results, however, much lower ductility recovery temperatures,
were found in samples with large grain size and a small amount of δ-phase. For re-melted
samples several hundred degrees lower recovery temperatures were measured.
351
Introduction
The approximately 5 % niobium in alloy 718 enters as the principal hardening element into the
γ” precipitates, with the chemical composition Ni3Nb and with a body centered tetragonal (BCT)
type of crystal structure, coherent with the face centered cubic (FCC) alloy matrix. The BCT
phase has an inherent mismatch and the associated strain fields around the precipitates add
considerably to the strengthening of the alloy. However, a thermal instability factor is inherent in
the incoherence and alloy 718 is not recommended for high stress applications at temperatures
above 650 °C. This is one of the fundamental differences to the γ’ type of superalloys where the
γ’ precipitates are of an ordered FFC type and with much less mismatch with the FCC matrix [1].
At temperatures slightly above the upper recommended temperature the δ-phase, of body
centered tetragonal (BCT) crystal structure, precipitates and growths into platelets [2]. The
chemical composition of the γ” and the δ-phase is identical, Ni3Nb, although with the different
crystal structures and the latter can thus not precipitate without the dissolution of former - the
δ−phase is the stable one. Depending on the actual Nb content within the specification range,
4.75 – 5.5 % by weight, the solution temperature for the δ-phase varies from approximately 850
°C to 1050 °C. With current Nb levels of wrought material today, with typically more than 5.2
%, δ is still stable at 1000 °C.
Since the δ−phase reduces the amount of Nb available for γ” hardening and the amount has to be
limited. Still a certain level is necessary for the grain size control in wrought material through its
pinning effect during the manufacturing sequences. As a consequence, at solution heat
treatments, prior to hardening, the temperature should not be so high that all δ-phase dissolves,
e.g. 1050 °C, and temperatures around 950 °C are often used. At such temperatures, however, δphase is not only stable but it also continues to precipitate at the typical Nb-levels mentioned.
For each repair cycle involving welding on structural parts a solution and ageing process is
needed to restore the full properties and the δ-phase level increases with each cycle and may
finally be unacceptable.
There are several reasons for limiting the acceptable amount of δ-phase and first of all because it
drains Nb from the matrix and reduces the strength achievable through the age hardening.
Secondly, it is less ductile than the matrix which may be readily understood from the BCT
structure, but, contrary to what the plate like morphological similarities with the topologically
closed packed (TCP) phases of the γ’ alloys may suggest, it is far from as brittle and it adheres
well to the matrix allowing for significant deformation before the platelets separates from the
matrix [3].
A third reason for limiting the amount of δ-phase relates to its possible deteriorating effects on
weldability [4] which is the purpose of the present paper to shed light on. In heavily constrained
areas the reduced strain at fracture of material with high δ-phase levels may lead to hot shortness
and weld cracks. Of more fundamental concern has been possible incipient melting caused by
preferential nucleation and precipitation of the Nb-rich δ-phase at grain boundaries.
352
Experimental
Cast material
Practical TIG weld experiments were carried out on hollow vanes excised from a retired large
cast gas turbine structure under heavily constrained conditions, as illustrated in figure 1. Due to
its inherent inhomogeneity cast alloy 718 is more susceptible to weld cracking than the wrought
material. The actual chemical composition is given in table I together with that of the wrought
material used for the hot tensile testing.
Table I. Chemical analysis in weight % of the tested alloy 718 materials.
Structural Casting
Wrought
Ni Cr Fe Mo Nb Al Ti C
B
P
S
Ni
Bal. 18.9 18.3 2.94 4.93 0.46 0.94 0.05 0.004 0.009 0.004 Bal.
Bal. 17.7 18.3 2.88 5.12 0.50 0.97 0.02 0.003 0.006 <0.003 Bal.
Fixture
Vane
Fixture
Figure 1 Heavily constrained welding arrangement with the vane welded into the fixture device
before actual welding trials. To the right a close up view on a disk shaped piece of material TIG
welded into a corresponding circular hole in the vane
As the purpose of this investigation is to evaluate the effect of δ-phase, comparison was made of
the material with different levels of this phase. A high level in this cast material was achieved
after precipitation at 954 °C during 30 hours. With an additional heat treatment at 1040 °C for 2
hours virtually all of this δ−phase was dissolved as seen in figure 2. The uneven distribution of
the phase reflects the strong tendency for segregation of the niobium in the casting process to
locations in the vicinity of grain boundaries where solidification terminates.
353
0.1 mm
0.1 mm
Figure 2 Preferential precipitation of δ-phase at grain boundaries in the cast material after 30
hours at 954 °C (left) and followed by a 2 hours solution at 1040 °C (right).
Wrought material
The wrought material for hot tensile testing was heat treated into a fully solutioned, δ-free, state
and into a high level δ-phase state both with two different grain sizes and also into a large grain
state with a limited amount of δ-phase as shown in table II.
Table II. Heat treatments and microstructure before hot tensile testing.
Acronym
Heat treatment
Microstructure after heat treatment
LG
1100 °C 1h, furnace cooling
Homogenous γ, GS 150-200µm
SG
(Small Grain)
1040 °C 1h , water quench
Homogenous γ, GS 50-75µm
SGD
(Small Grain Delta)
1040 °C 1h, water quench,
900°C 24h, water quench
15% δ-phase, GS 50-75µm.
LGD
(Large Grain Delta)
1100 °C 1h, water quench,
900 °C 24h, water quench
12% δ-phase δ, GS 150-200µm.
LGLD
(Large Grain Limited Delta)
1100 °C 1h, furnace cooling,
960 °C 24h, furnace cooling
2% coarse δ-platelets, frequently at
grain boundaries, GS 150-200µm
In-situ melted
none
As in the delivery condition
(Large Grain)
(Carbides and nitrides are present in all specimens and not affected by any heat treatments except
for the in situ melted samples)
354
The microstructure of the small grain (SG) specimens without δ-phase shown in figure 3a also
illustrates the microstructure of the large grain (LG) specimens. The δ-phase level shown in
figure 3b of the small grain δ-phase (SGD) specimens similarly illustrates the level of the large
grain δ (LGD) specimens. The microstructure of the large grain material with limited amount of
δ-phase (LGLD) is shown in figure 3c.
0.1 mm
a
b
0.05 mm
c
0.2 mm
Figure 3. Typical microstructures of the examined wrought material after specific heat
treatments. a) Fully solutioned after 1 hour at 1040 °C (SG). b) Same material as in fig a but with
an additional 900 °C 24 hours δ-phase precipitation heat treatmen (SGD). c) Grain growth at
1100 °C for 1 hour followed by limited δ-phase precipitation at 960 °C during 24 hours
(LGLD).
The hot ductility testing was performed in an equipment especially developed for the study of
strength and ductility during melting and solidification of metals at the Department of Casting
Technology at The Royal Institute of Technology in Stockholm, Sweden, [5] schematically
shown in figure 4.
355
1.
2.
3.
4.
5.
6.
Specimen with hole for TC
Grip
Tension rod
Quartz tube
Seal
Elliptic focusing reflectors
Figure 4. Hot tensile test equipment and specimen geometry.
Due to the small dimensions of the specimens, 4 mm diameter and 40 mm length, and a good
focusing power of the elliptic 1 kW mirror furnace a heating rate of 300 °C/min as well as a
cooling rate of 400 °C/min was possible to use in our experiments. For the same reasons it was
possible to melt a limited amount of material at the center of the specimen with an approximate
length of the heated zone of 5 mm , a droplet melt which was kept in position by surface tension
and thus enabled the study of the full temperature range of the welding process. The temperature
measurements were made by a (Pt-PtRh) thermocouple located in a hole centered in each
specimen. During the testing, the specimens were located inside a quartz tube and protected by a
flow of pure argon gas.
The span between the temperature where the material loses its strength completely on heating,
the zero strength temperature (ZST), and the temperature when the material has recovered the
ductility during cooling, the ductility recovery temperature (DRT) was used as a measure of the
susceptibility to cracking in the HAZ . Such a temperature span is logical to use in the sense that
the larger this interval, with the material in a ‘brittle’ state, the larger is the risk for cracks to
356
develop in the heat affected zone of the material when subjected to strain during the cooling
down phase of the welding process.
The heating and cooling cycle at testing is schematically illustrated in figure 5. The specimens
were heated to 10-20°C below the zero strength temperature, ZST, determined for each type of
material by heating the sample at rate of 5°C/ s with an applied load of 50N, until the samples
fractured. The in situ solidified specimens were for several practical reasons tested according to
another cycle as also seen in figure 5.
In situ solidified
Reheated
Figure 5. Schematic heating and cooling cycle at testing of reheated specimens with specific
microstructures to the left and the cycle used for the in situ solidified specimens to the right
The strain rate during the testing was set by the 0.5 mm/ s tension rod rate, which corresponds to
an approximate strain rate of 0.1 s-1, assuming that the length of the deformed zone was the same
as the length of the heated zone, i.e. 5 mm.
Evaluation of microstructures and fracture surfaces were made by standard metallographic
techniques including SEM and EDS.
Results
Practical weld experiments on cast material
As mentioned it was very difficult to provoke any cracking in the heavily constrained weld
experiments on the ex-service cast vanes. Nor by penetrant inspection neither by x-ray
examination was it possible to detect any cracking. However, after extensive successive
metallographic examination one crack in the vane without δ-phase and one crack in the vane
with δ-phase were possible to detect in the HAZ adjacent to the weld material, both in the
millimetre size region.
357
Hot tensile testing
The ZST for the different materials was determined by heating the sample at a rate of 5 °C/ s
with an applied load of 50N, until the samples fractured. For the two materials containing large
amount of δ-phase ZST was found to be 1223 °C and for the material without δ-phase (including
the material with the very limited amount) it was found to be 1234 °C while for the in situ melted
material the ZST was established to be 1238 °C.
During the ductility recovery tensile testing the peak temperature was chosen 10 – 20 °C below
ZST before the cooling down and testing at the specified testing temperatures as indicated in
figure 5.
The actual ductilities measured by the reduction of area for the different materials are shown in
figure 6. Ductility for the materials with high amounts of δ-phase (figure 3 b) and without any δphase is rapidly recovered and reach 100 %, all within 100 °C from the ZST. However, the
material with the large grains with a limited amount of δ does show a significantly lower
temperature for the return of the ductility, actually at a more than 100 °C lower temperature.
Finally, in the in situ solidified material there is a limited ductility which arrives first at very low
temperatures.
Area reduction-Temperature on Cooling
100
Area reduction [%]
80
Small Grain
Small Grain Delta
Large Grain
Large Grain Delta
Large Grain Lim Delta
In Situ Solidified
60
40
20
ZST (on heating)
0
900
950
1000
1050
1100
Temperature [°C]
1150
1200
1250
Ts (on heating DTA)
Figure 6. Ductility of the different states of the wrought material after testing at specified
temperatures on cooling from the zero strength temperature and measured as area reduction.
For the in situ solidified material and the small grain material with and without δ-phase the
gradual return of the ultimate tensile when temperature decreased were, at what first may seem
surprisingly, similar as shown in figure 7. However, at the very high temperature testing the yield
or rather creep stress is basically determined by the resistance to dislocation mobility through by
358
the chemistry and with the microstructural features like grain size having a secondary influence.
Ultimate stress is reached instantaneously considering the very limited amount of material
involved (4 mm diameter x 5 mm heated length) followed by either necking or brittle fracture.
Maxium Force on Cooling
3500
3000
Maximum force [N]
2500
2000
Small Grain
Small Grain Delta
In Situ Solidified
1500
1000
500
0
900
950
1000
1050
1100
1150
1200
1250
1300
Temperature [°C]
Figure 7. Maximum force registered during the tensile testing at the selected temperatures after
cooling from the zero strength temperature.
Typical appearances of fractured specimens at full ductility and at zero ductility are shown in
figure 8. In the figure the material with large grains and without δ-phase, tested at 1050 °C and at
1162 °C , respectively, are used for illustration.
Figure 8. Typical appearances of fractured specimens with full ductility and with no ductility at
the testing. The fracture surfaces of the material with large grains and without δ-phase tested at
1050 °C with 100 % ductility and at 1162 °C with 0 % are shown to the left and to the right,
respectively, are here used for illustration. In the ductile specimen the thermocouple, diameter 1
mm, remains in the position of the testing while it has been pulled out from the brittle specimen
after the testing.
359
Discussion
The results from the practical weld experiments on cast material as shown in figure 1 confirms
the excellent weldability of alloy 718. Since the niobium is very strongly segregated to the grain
boundaries in the cast material the precipitated δ-phase also accumulates to the vicinity of the
grain boundaries, as shown in figure 2. Thus, any deterioration effect of the δ-phase on
weldability involving intergranular fracture in the HAZ must be more pronounced in cast
material in comparison with wrought material with the same total amount of δ-phase present
since the niobium segregation is much less pronounced in the wrought material where the much
finer grain structure adds to the crack resistance. From this rational we may draw the conclusion,
although no welding experiments were conducted, that wrought material should be able to
sustain much more δ-phase than cast material without causing any weldability problems.
The results from the tensile testing on the wrought material support this conclusion. Looking first
into the group of four materials containing the large grain and small grain material without δphase and the same two materials with a large amount of δ they all recover the ductility within
100 °C from the zero strength temperature and this indicates a good weldability.
Let us now consider the details of the tensile test results. The ductility of the fine grain material
without δ-phase δ returns at significantly higher temperatures when compared with the same but
large grain size material and this reflects the well known better weldability of materials with a
fine grain size. Ductility returns slightly more gradually in the materials containing large amount
of δ-phase as compared with the sluggish return in the materials without δ-phase. It also seems
as if the δ−levels out the effect of the grain size. A conclusion may here be that the presence of
δ-phase does not add to the deterioration of the weldability already induced by an increased grain
size, but data is limited. Another conclusion that may be drawn is that small grain material with
δ-phase is worse from a welding point of view than the large grain material with δ-phase but
here the data are even scarcer and in addition it is difficult to find a rational explanation that
supports such a conclusion.
The part of the tensile test results which we find most difficult to explain is the ductility return
curve for the material where the δ-phase precipitation temperature was 960 °C instead of 900 °C
producing a very limited, but in comparison, very coarse δ-phase at the grain boundaries as seen
in figure 3. If the amount of δ can be considered to have a detrimental effect on the ductility
recovery temperature one would have expected to see the material with a higher δ-phase content
to have the ductility return at a lower temperature. Now we see the opposite. Although we are
not able to draw any final conclusions from these anomalous results the metallographic
examination of the fractured test bars may shed some light on the issue.
In figure 9 features of the brittle fracture surface of the material with the limited amount of δ are
shown from a specimen tested at 1150 °C. In the low magnification (9a) the intergranular
fracture character is obvious with secondary cracks clearly visible. What may be the explanation
for the anomalous behaviour of this material is a conclusion that can be drawn from the higher
magnification micrograph 9b. It is here evident from the dark patches (two larger pointed out by
the white arrows) that a liquid phase with a good wetting capacityty was present during the
fracture process. Although it was not possible to analyse by EDS the dark appearance,
presumably given by the atomic number contrast, indicates that atoms of low atomic number
constitute a significant part of the composition. In these circumstances sulphur or boron may be
candidates – both eluding EDS analysis and both capable of producing very low melting
360
compounds and known to have a deleterious influence on weldability. However, we have no
explanation to why such phases would appear during the 24 hour heat treatment at 960 °C but
neither at a higher (1 hour at 1040 °C) nor at a lower (24 hour at 900 °C) temperature. In the
other high magnification micrograph (9c) indications of another liquid phase of lower contrast is
seen. What is seen here may be traces of what was the basic concern of our investigation and the
suggestions in the literature that δ-phase at grain boundaries during the welding process may
supply niobium for constitutional liquation through the Laves eutectic. In figure 10c traces of the
thinner δ-platelets of the small grain material can still be seen after the tensile testing. This
observation supports the possibility that the niobium rich area which remains after the fast
solution of the thin δ-platelets may give enrichment for such a constitutional Laves phase
eutectic formation. The coarser platelets precipitated at 960 °C and at the grain boundaries
should be even more powerful in this respect.
a
0.2 mm
b
0.02 mm
c
0.1 mm
Figure 9. Features of the brittle fracture surface of the material with the limited amount of δphase from the specimen tested at 1150 °C. a) Intergranular fracture surface with secondary
cracks. b) Traces of wetting liquid with low atomic number element contrast. c) Traces of
possible high niobium melt.
For comparison, the fracture surface from a brittle fracture of the small grain material (1 hour
solution at 1040 °C) with a large amount of δ-phase (24 hour at 900 °C) is shown in figure 10.
The specimen was tested at 1075 °C with zero ductility. In the lower magnification (10a) the
intergranular character of the fracture is evident but much less pronounced than for the
anomalous material shown in figure 9. The irregular topography of the surface (10a and b)
indicates a certain local deformation during the fracture process as does the secondary cracks
below the fracture surface (10c). The grain boundary cohesion is here stronger but not so strong
as to allow for any measurable overall plastic deformation before fracture. At the fracture surface
there are also a few isolated patches with evident low grain boundary cohesion(10b). In our
investigation of this material it was however not possible to attach any cause to these features. In
all, our results stresses the importance of understanding the grain boundary chemistry and
reactions involved in all the temperature exposure of super alloys prior and during the welding
process. Of special importance are grain boundary active elements like sulphur, phosphorus and
boron and segregation and partitioning in connection with secondary phases – complex relations
which must be in focus when optimizing weldability of 718 type and other types of superalloys.
361
a
0.2 mm
0.02 mm
b
c
0.02 mm
Figure 10. The brittle fracture surface (10a and 10b) and a cross section close to the fracture
surface (10c) of the small grain material (1 hour solution at 1040 °C) with a large amount of δphase (24 hour at 900 °C) tested at 1075 °C with zero ductility. a) Intergranular fracture with
micro-level grain boundary ductility. b) Isolated patch of low grain boundary coherence in
micro-ductile surroundings. c) Grain structure close to the fracture surface with a secondary
crack and traces of solutioned δ-platelets
Conclusions
•
The presence of δ-phase has minimal if any effect on the weldability of alloy 718.
•
Anomalous, low hot ductility, behaviour of wrought material after precipitation of δphase at 960 °C for 24 hours is associated with the occurrence of a liquid with good
ability to wet grain boundaries.
Acknowledgements
KME, Sweden supported the major part of the work in this paper. Special thanks to coop student
Robert Söderberg for the work involving the weldability of the cast ex-service material.
References
1 C.T. Sims and W.C. Hagel, “The Superalloys, Wiley, New York, 1972, 171-178.
2 S. Azadian, ”Aspects of Precipitation in the Alloy Inconel 718” (Ph.D. thesis, Luleå University
of technology, Sweden, 2004), 71-83.
3 G. Sjöberg and N-G. Ingesten, “Grain Boundary δ-phase Morphologies, Carbides and
Notch Rupture Sensitivity of Cast Alloy 718”, Superalloys 718, 625, 706 and Various
Derivatives, Loria, E. A. ed., (The Minerals, Metals & Materials Society, 1991), 603-620.
4 M.E. Mehl and J.C. Lippold, “Effect of δ-phase Precipitation on the Repair Weldability of
Alloy 718”, Superalloys 718, 625 ,706 and Various Derivatives, Edited by E.A. Loria,
(The Minerals, Metals & Materials Society, 1997), 731 -741.
5 Tomas Antonsson, ”On the Interaction Between Liquid/Solid During Sintering and
Solidification” (Ph.D. thesis, Royal Institute of Technology, Sweden, 2003), Suplement 8.
362